CUSTOMER SEGMENTATION - USING K-MEAN CLUSTERING IN PYTHON¶

Importing Libraries

In [3]:
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
import os
import warnings
%matplotlib inline
import plotly.express as px
from sklearn.cluster import KMeans
from sklearn.preprocessing import LabelEncoder
import plotly.io as pio
pio.renderers.default='notebook'
In [ ]:

Importing Dataset

In [5]:
df = pd.read_csv(r"G:\2025 DS FOLDER\CS - Customer Segmentation\Kmean Clustering in Python\Mall_Customers.csv")
In [6]:
df.head(10)
Out[6]:
CustomerID Gender Age Annual Income (k$) Spending Score (1-100)
0 1 Male 19 15 39
1 2 Male 21 15 81
2 3 Female 20 16 6
3 4 Female 23 16 77
4 5 Female 31 17 40
5 6 Female 22 17 76
6 7 Female 35 18 6
7 8 Female 23 18 94
8 9 Male 64 19 3
9 10 Female 30 19 72
In [7]:
df.shape
Out[7]:
(200, 5)
In [8]:
df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 200 entries, 0 to 199
Data columns (total 5 columns):
 #   Column                  Non-Null Count  Dtype 
---  ------                  --------------  ----- 
 0   CustomerID              200 non-null    int64 
 1   Gender                  200 non-null    object
 2   Age                     200 non-null    int64 
 3   Annual Income (k$)      200 non-null    int64 
 4   Spending Score (1-100)  200 non-null    int64 
dtypes: int64(4), object(1)
memory usage: 7.9+ KB

There are no null values in the dataset. The data is clean and ready for use.

In [10]:
df.describe()
Out[10]:
CustomerID Age Annual Income (k$) Spending Score (1-100)
count 200.000000 200.000000 200.000000 200.000000
mean 100.500000 38.850000 60.560000 50.200000
std 57.879185 13.969007 26.264721 25.823522
min 1.000000 18.000000 15.000000 1.000000
25% 50.750000 28.750000 41.500000 34.750000
50% 100.500000 36.000000 61.500000 50.000000
75% 150.250000 49.000000 78.000000 73.000000
max 200.000000 70.000000 137.000000 99.000000
In [11]:
#Renaming the columns
df.columns=['CustomerID', 'Gender', 'Age', 'Annual Income','Spending Score']
In [12]:
df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 200 entries, 0 to 199
Data columns (total 5 columns):
 #   Column          Non-Null Count  Dtype 
---  ------          --------------  ----- 
 0   CustomerID      200 non-null    int64 
 1   Gender          200 non-null    object
 2   Age             200 non-null    int64 
 3   Annual Income   200 non-null    int64 
 4   Spending Score  200 non-null    int64 
dtypes: int64(4), object(1)
memory usage: 7.9+ KB

The column names successfully changed

Peforming EDA - Exporatory Data Analysis

In [15]:
plt.figure(figsize=(18,6))

plt.subplot(1,3,1)
sns.histplot(df['Age'],color = 'skyblue',kde = True)
plt.title('Customer Age', fontsize = 16)
plt.xlabel('Customer Age', fontsize = 14)
plt.ylabel('Frequency', fontsize = 14)


plt.subplot(1,3,2)
sns.histplot(df['Annual Income'],color = 'forestgreen', kde = True)
plt.title('Customer Annual Incoome', fontsize = 16)
plt.xlabel('Income', fontsize = 14)


plt.subplot(1,3,3)
sns.histplot(df['Spending Score'],color = 'royalblue', kde = True)
plt.title('Customer Spending Score', fontsize = 16)
plt.xlabel('Spending Score', fontsize = 14)


plt.show()
No description has been provided for this image
In [17]:
plt.figure(figsize=(8,6))
sns.countplot(x='Gender', hue = 'Gender', data=df, palette='winter', legend=False)
plt.title('Count Plot of Gender', fontsize=16)
plt.xlabel('Gender', fontsize=14)
plt.ylabel('Count', fontsize=14)
plt.xticks([0,1],['Male','Female'],fontsize=12 )
plt.show()
No description has been provided for this image
In [18]:
sns.set_style('white')
new=sns.PairGrid(df.drop(['CustomerID', 'Gender'],axis=1))
new.map(sns.regplot, color='teal')
plt.suptitle('Relationship between Age, Annual Income and Spending Score',y=1.05,fontsize=14)
plt.show()
No description has been provided for this image
In [19]:
sns.lmplot(x='Age', y='Annual Income', data=df,fit_reg=False, hue='Gender', height=6, palette='plasma')
plt.title('Relationship between Age, Annual Income and Gender', fontsize=14)
plt.xlabel('Age', fontsize=13)
plt.ylabel('Annual Income', fontsize=13)
plt.show()
No description has been provided for this image
In [20]:
sns.lmplot(x='Annual Income', y='Spending Score', data=df,fit_reg=False, hue='Gender', height=6, palette='Dark2')
plt.title('Annual Income vs Spending Score vs Gender', fontsize=14)
plt.xlabel('Annual Income', fontsize=13)
plt.ylabel('Spending Score', fontsize=13)
plt.show()
No description has been provided for this image

KMean Clustering Segmentation Based on Age and Spending Score, Finding the Optimal Number of Clusters First, we find the optimal number of clusters by elbow method.

In [22]:
#SETTING THE ENVIRONMENT VARIABLE TO AVOID MEMORY LEAK
os.environ['OMP_NUM_THREADS'] = '1'
# Suppress the specific warning from sklearn
warnings.filterwarnings("ignore", category=UserWarning, module='sklearn.cluster._kmeans')
#Kmean Clustering
X1=df[['Age','Spending Score']]
inertia1=[]
for n in range(1 , 11):
    model1=KMeans(n_clusters=n, init='k-means++', n_init=10, max_iter=300, tol=0.0001, random_state=111,algorithm='lloyd')
    model1.fit(X1)
    inertia1.append(model1.inertia_)
print(inertia1)

[171535.5, 75949.15601023019, 45840.67661610867, 28165.58356662934, 23818.84825158133, 19510.328802776174, 15514.193134351035, 13012.800984809695, 11513.134979462546, 10590.814352780133]
In [23]:
plt.figure(figsize=(8,6))
plt.plot(list(range(1 , 11)), inertia1, color='royalblue', marker='o',linewidth=2, markersize=12, markerfacecolor='m', markeredgecolor='m')
plt.title('Inertia vs. Number of Clusters',fontsize=18)
plt.xlabel('Number of Clusters',fontsize=15)
plt.ylabel('Inertia',fontsize=15)
plt.show()
No description has been provided for this image

We can see that if the number of clusters is smaller than 4, the inertia has a high value but if the number of clusters is larger than 4, the inertia is relatively constant. So we choose 4 as the optimal number of clusters.

In [25]:
#Performing Clustering with the Optimal Number of Clusters - 4
model1=KMeans(n_clusters=4, init='k-means++', n_init=10, max_iter=300, tol=0.0001, random_state=111,algorithm='lloyd')
model1.fit(X1)
labels1=model1.labels_
centroids1=model1.cluster_centers_ 
# Suppress the specific warning from sklearn
warnings.filterwarnings("ignore", category=UserWarning, module='sklearn.cluster._kmeans')

Visualizing the Cluster/segmentation

In [27]:
plt.figure(figsize=(11,6))
sns.set_style('white')
plt.scatter(x=df['Age'], y=df['Spending Score'],c=labels1,cmap='winter')
plt.scatter(x = centroids1[: , 0] , y =  centroids1[: , 1], s = 300 , c = 'red')
plt.xlabel('Age',fontsize=15)
plt.ylabel('Spending Score',fontsize=15)
plt.title('Customer Segmentation Based on Age and Spending Score',fontsize=18)
plt.show()
No description has been provided for this image

Segmentation Based on Annual Income and Spending Score Finding the Optimal Number of Clusters

In [29]:
X2=df[['Annual Income','Spending Score']]
inertia2=[]
for n in range(1 , 11):
    model2=KMeans(n_clusters=n, init='k-means++', n_init=10, max_iter=300, tol=0.0001, random_state=111,algorithm='lloyd')
    model2.fit(X2)
    inertia2.append(model2.inertia_)
print(inertia2)

# Suppress the specific warning from sklearn
warnings.filterwarnings("ignore", category=UserWarning, module='sklearn.cluster._kmeans')

[269981.28, 181363.59595959596, 106348.37306211119, 73679.78903948834, 44448.45544793371, 37265.86520484346, 30241.343617936585, 25022.485004530354, 21841.97825674864, 19707.258979794773]

Then Lets Plot to see the Point of optimal number cluster

In [31]:
plt.figure(figsize=(8,6))
sns.set_style('white')
plt.plot(list(range(1 , 11)), inertia2, color='forestgreen', marker='o',linewidth=2, markersize=12, markerfacecolor='c', markeredgecolor='c')
plt.title('Inertia vs. Number of Clusters',fontsize=18)
plt.xlabel('Number of Clusters',fontsize=15)
plt.ylabel('Inertia',fontsize=15)
plt.show()
No description has been provided for this image

We can see that if the number of clusters is smaller than 5, the inertia has a high value but if the number of clusters is larger than 5, the inertia is relatively constant. So we choose 5 as the optimal number of clusters.

Performing Clustering with the Optimal Number of Clusters

In [34]:
#Performing Clustering with the Optimal Number of Clusters - 5 (ANNUAL INCOME VS SPENDING SCORE)
model2=KMeans(n_clusters=5, init='k-means++', n_init=10, max_iter=300, tol=0.0001, random_state=111,algorithm='lloyd')
model2.fit(X2)
labels2=model2.labels_
centroids2=model2.cluster_centers_
# Suppress the specific warning from sklearn
warnings.filterwarnings("ignore", category=UserWarning, module='sklearn.cluster._kmeans')

Visualizing the CLuster

In [36]:
plt.figure(figsize=(11,6))
sns.set_style('white')
plt.scatter(x=df['Annual Income'], y=df['Spending Score'],c=labels2,cmap='winter')
plt.scatter(centroids2[:,0], centroids2[:,1],c='red',s=200)
plt.xlabel('Annual Income',fontsize=15)
plt.ylabel('Spending Score',fontsize=15)
plt.title('Customer Segmentation Based on Annual Income and Spending Score',fontsize=18)
plt.show()
No description has been provided for this image

Performing Clustering Based on Age, Annual Income and Spending Score

The Determining The Optimal Number of Clusters Based on The Condition Above

In [39]:
X3=df[['Age','Annual Income','Spending Score']]
inertia3=[]
for n in range(1 , 10):
    model3=KMeans(n_clusters=n, init='k-means++', n_init=10, max_iter=300, tol=0.0001, random_state=111,algorithm='lloyd')
    model3.fit(X3)
    inertia3.append(model3.inertia_)
print(inertia3)
# Suppress the specific warning from sklearn
warnings.filterwarnings("ignore", category=UserWarning, module='sklearn.cluster._kmeans')

[308812.78, 212840.1698209719, 143342.751571706, 104366.15145556198, 75479.76429338778, 58300.44332159069, 51525.773770373766, 44307.87341670445, 40932.6282276547]

Plotting the Inertia against the Number of Cluster to View the number of Clusters

In [41]:
plt.figure(figsize=(8,6))
plt.plot(list(range(1 , 10)), inertia3, color='royalblue', marker='o',linewidth=2, markersize=12, markerfacecolor='violet', markeredgecolor='violet')
plt.title('Inertia vs. Number of Clusters',fontsize=18)
plt.xlabel('Number of Clusters',fontsize=15)
plt.ylabel('Inertia',fontsize=15)
plt.show()
No description has been provided for this image

We can see that if the number of clusters is smaller than 6, the inertia has a high value but if the number of clusters is larger than 6, the inertia is relatively constant. So we choose 6 as the optimal number of clusters.

In [43]:
#Performing Clustering with the Optimal Number of Clusters - 6
model3=KMeans(n_clusters=6, init='k-means++', n_init=10, max_iter=300, tol=0.0001, random_state=111,algorithm='lloyd')
model3.fit(X3)
inertia3.append(model3.inertia_)
labels3=model3.labels_
centroids3=model3.cluster_centers_
# Suppress the specific warning from sklearn
warnings.filterwarnings("ignore", category=UserWarning, module='sklearn.cluster._kmeans')
In [44]:
#Plotting the Cluster
#This plot will reguire 3D plotting.
fig=px.scatter_3d(data_frame=df,x='Age',y='Annual Income',z='Spending Score',color=labels3, color_continuous_scale='bluered')
fig.update_layout(
    title={
        'text': 'Customer Segmentation Based on Age, Annual Income and Spending Score',
        'y':0.95,
        'x':0.5,
        'xanchor': 'center',
        'yanchor': 'top'})
fig.show()
In [ ]: